control_earl {polle}

Control arguments for Efficient Augmentation and Relaxation Learning

Description

control_earl sets the default control arguments for efficient augmentation and relaxation learning , type = "earl". The arguments are passed directly to DynTxRegime::earl() if not specified otherwise.

Usage

control_earl(
  moPropen,
  moMain,
  moCont,
  regime,
  iter = 0L,
  fSet = NULL,
  lambdas = 0.5,
  cvFolds = 0L,
  surrogate = "hinge",
  kernel = "linear",
  kparam = NULL,
  verbose = 0L
)

Arguments

moPropen

Propensity model of class "ModelObj", see modelObj::modelObj.

moMain

Main effects outcome model of class "ModelObj".

moCont

Contrast outcome model of class "ModelObj".

regime

An object of class formula specifying the design of the policy/regime.

iter

Maximum number of iterations for outcome regression.

fSet

A function or NULL defining subset structure.

lambdas

Numeric or numeric vector. Penalty parameter.

cvFolds

Integer. Number of folds for cross-validation of the parameters.

surrogate

The surrogate 0-1 loss function. The options are "logit", "exp", "hinge", "sqhinge", "huber".

kernel

The options are "linear", "poly", "radial".

kparam

Numeric. Kernel parameter

verbose

Integer.

Value

list of (default) control arguments.